[nieve]

/snark I've certainly had PostgreSQL crash and all it took was someone in a Verizon datacenter accidentally yanking out the cables connecting the server to its RAID arrays. It even crashed a second time when someone "tripped over a cable" and took out the arrays again and a third time when the redundant power supplies turned out not to be...

OTOH complete loss of storage resulted in sub-second data loss with autorecovery the first two times using a five year old Pg installation and minimal issues the third time. The master server had a three year uptime on the main server process when it went down. Overall I'm pretty comfortable with that track record.